Inference Algorithms

Classes
class	opengm::AlphaBetaSwap< GM, INF >
	Alpha-Beta-Swap Algorithm. More...
class	opengm::AlphaExpansion< GM, INF >
	Alpha-Expansion Algorithm. More...
class	opengm::AlphaExpansionFusion< GM, ACC >
	Alpha-Expansion-Fusion Algorithm uses the code of Alexander Fix to reduce the higer order moves to binary pairwise problems which are solved by QPBO as described in Alexander Fix, Artinan Gruber, Endre Boros, Ramin Zabih: A Graph Cut Algorithm for Higher Order Markov Random Fields, ICCV 2011. More...
class	opengm::AStar< GM, ACC >
	A star search algorithm. More...
class	opengm::PrimalLPBound< GM, ACC >
	[class primallpbound] PrimalLPBound - estimating primal local polytope bound and feasible primal solution for the local polytope relaxation of the MRF energy minimization problem Based on the paper: B. Savchynskyy, S. Schmidt Getting Feasible Variable Estimates From Infeasible Ones: MRF Local Polytope Study. arXiv:1210.4081 Submitted Oct. 2012 More...
class	opengm::Bruteforce< GM, ACC >
	Brute force inference algorithm. More...
class	opengm::CombiLP< GM, ACC, LPSOLVER >
	CombiLP Savchynskyy, B. and Kappes, J. H. and Swoboda, P. and Schnoerr, C.: "Global MAP-Optimality by Shrinking the Combinatorial Search Area with Convex Relaxation". In NIPS, 2013. More...
class	opengm::DualDecompositionSubGradient< GM, INF, DUALBLOCK >
	Dual-Decomposition-Subgradient Inference based on dual decomposition using sub-gradient descent Reference: Kappes, J. H. and Savchynskyy, B. and Schnoerr, C.: "A Bundle Approach To Efficient MAP-Inference by Lagrangian Relaxation". In CVPR 2012, 2012. More...
class	opengm::DynamicProgramming< GM, ACC >
	DynamicProgramming. More...
class	opengm::external::AD3Inf< GM, ACC >
	AD3 . More...
class	opengm::external::DAOOPT< GM >
	DAOOPT DAOOPT inference algorithm class. More...
class	opengm::external::FastPD< GM >
	FastPD FastPD inference algorithm class. More...
class	opengm::external::GCOLIB< GM >
	GCOLIB GCOLIB inference algorithm class. More...
class	opengm::external::GRANTE< GM >
	GRANTE GRANTE inference algorithm class. More...
class	opengm::external::libdai::MeanField< GM, ACC >
	Mean Field : [?]. More...
class	opengm::external::libdai::TreeExpectationPropagation< GM, ACC >
	Tree Expectation Propagation : [?]. More...
class	opengm::external::libdai::TreeReweightedBp< GM, ACC >
	tree reweighted belief propagation : [?] More...
class	opengm::external::MPLP< GM >
	MPLP MPLP inference algorithm class. More...
class	opengm::external::MRFLIB< GM >
	MRFLIB MRFLIB inference algorithm class. More...
class	opengm::external::QPBO< GM >
	QPBO Algorithm. More...
class	opengm::external::TRWS< GM >
	message passing (BPS, TRWS): [?] More...
class	opengm::GibbsMarginalVisitor< GIBBS >
	Visitor for the Gibbs sampler to compute arbitrary marginal probabilities. More...
class	opengm::GraphCut< GM, ACC, MINSTCUT >
	A framework for min st-cut algorithms. More...
class	opengm::GreedyGremlin< GM, ACC >
	GREEDY GREMLIN. More...
class	opengm::HQPBO< GM, ACC >
	HQPBO Algorithm . More...
class	opengm::ICM< GM, ACC >
	Iterated Conditional Modes Algorithm J. E. Besag, "On the Statistical Analysis of Dirty Pictures", Journal of the Royal Statistical Society, Series B 48(3):259-302, 1986. More...
class	opengm::InfAndFlip< GM, ACC, INF >
	Inference and Flip . More...
class	opengm::LazyFlipper< GM, ACC >
	A generalization of ICM B. Andres, J. H. Kappes, U. Koethe and Hamprecht F. A., The Lazy Flipper: MAP Inference in Higher-Order Graphical Models by Depth-limited Exhaustive Search, Technical Report, 2010, http://arxiv.org/abs/1009.4102. More...
class	opengm::LOC< GM, ACC >
	LOC Algorithm K. Jung, P. Kohli and D. Shah, "Local Rules for Global MAP: When Do They Work?", NIPS 2009. More...
class	opengm::LPCplex< GM, ACC >
	Optimization by Linear Programming (LP) or Integer LP using IBM ILOG CPLEX http://www.ilog.com/products/cplex/. More...
class	opengm::LPGurobi< GM, ACC >
	Optimization by Linear Programming (LP) or Integer LP using Guroi http://www.gurobi.com. More...
struct	opengm::LSA_TR_WeightedEdge
	Local Submodular Approximation with Trust Region regularization . More...
class	opengm::MQPBO< GM, ACC >
	[class mqpbo] Multilabel QPBO (MQPBO) Implements the algorithms described in i) Ivan Kovtun: Partial Optimal Labeling Search for a NP-Hard Subclass of (max, +) Problems. DAGM-Symposium 2003 (part. opt. for potts) ii) P. Kohli, A. Shekhovtsov, C. Rother, V. Kolmogorov, and P. Torr: On partial optimality in multi-label MRFs, ICML 2008 (MQPBO) iii) P. Swoboda, B. Savchynskyy, J.H. Kappes, and C. Schnörr : Partial Optimality via Iterative Pruning for the Potts Model, SSVM 2013 (MQPBO with permutation sampling) More...
struct	opengm::ParamHeper
	Multicut Algorithm [1] J. Kappes, M. Speth, B. Andres, G. Reinelt and C. Schnoerr, "Globally Optimal Image Partitioning by Multicuts", EMMCVPR 2011 [2] J. Kappes, M. Speth, G. Reinelt and C. Schnoerr, "Higher-order Segmentation via Multicuts", Technical Report (http://ipa.iwr.uni-heidelberg.de/ipabib/Papers/kappes-2013-multicut.pdf) . More...
class	opengm::PartitionMove< GM, ACC >
	Partition Move Currently Partition Move only implements the Kernighan-Lin-Algorithm. More...
class	opengm::QPBO< GM, MIN_ST_CUT >
	QPBO Algorithm C. Rother, V. Kolmogorov, V. Lempitsky, and M. Szummer, "Optimizing binary MRFs via extended roof duality", CVPR 2007. More...
class	opengm::ReducedInference< GM, ACC, INF >
	[class reducedinference] Reduced Inference Implementation of the reduction techniques proposed in J.H. Kappes, M. Speth, G. Reinelt, and C. Schnörr: Towards Efficient and Exact MAP-Inference for Large Scale Discrete Computer Vision Problems via Combinatorial Optimization, CVPR 2013 More...
class	opengm::SAT< GM >
	2-SAT solver More...
class	opengm::SwendsenWang< GM, ACC >
	Generalized Swendsen-Wang sampling A. Barbu, S. Zhu, "Generalizing swendsen-wang to sampling arbitrary posterior probabilities", PAMI 27:1239-1253, 2005. More...
class	opengm::ADSal< GM, ACC >
	[class adsal] ADSal - adaptive diminishing smoothing algorithm Based on the paper: B. Savchynskyy, S. Schmidt, J. H. Kappes, C. Schnörr Efficient MRF Energy Minimization via Adaptive Diminishing Smoothing, In UAI, 2012, pp. 746-755 More...
class	opengm::TRWSi< GM, ACC >
	[class trwsi] TRWSi - tree-reweighted sequential message passing Based on the paper: V. Kolmogorov Convergent tree-reweighted message passing for energy minimization. IEEE Trans. on PAMI, 28(10):1568–1583, 2006. More...
class	opengm::LPCplex2< GM_TYPE, ACC_TYPE >
	LP inference with CPLEX. More...
class	opengm::LPGurobi2< GM_TYPE, ACC_TYPE >
	LP inference with Gurobi. More...

Functions
template<class PARAMETERS >
void	opengm::setSmoothingParametersForMarginals (PARAMETERS &params, size_t numIterations, typename PARAMETERS::ValueType temperature=1.0, typename PARAMETERS::Storage::StructureType decompositionType=PARAMETERS::Storage::GENERALSTRUCTURE)
	[function setSmoothingParametersForMarginals] setSmoothingParametersForMarginals - adjusts parameters of smoothing-based algorithms (NesterovAcceleratedGradient and ADSal) to obtain estimations of sum-prod margonals. More...

Detailed Description

Function Documentation

setSmoothingParametersForMarginals()

template<class PARAMETERS >

void opengm::setSmoothingParametersForMarginals	(	PARAMETERS &	params,
		size_t	numIterations,
		typename PARAMETERS::ValueType	temperature = 1.0,
		typename PARAMETERS::Storage::StructureType	decompositionType = PARAMETERS::Storage::GENERALSTRUCTURE
	)

[function setSmoothingParametersForMarginals] setSmoothingParametersForMarginals - adjusts parameters of smoothing-based algorithms (NesterovAcceleratedGradient and ADSal) to obtain estimations of sum-prod margonals.

For mathematical details see the papers: B. Savchynskyy, J. H. Kappes, S. Schmidt, C. Schnörr A Study of Nesterov's Scheme for Lagrangian Decomposition and MAP Labeling, in CVPR 2011 and B. Savchynskyy, S. Schmidt, J. H. Kappes, C. Schnörr Efficient MRF Energy Minimization via Adaptive Diminishing Smoothing, In UAI, 2012, pp. 746-755

Usage examples:

/* With NesterovAcceleratedGradient: */ NesterovAcceleratedGradient<GraphicalModelType,Minimizer>::Parameter params; setSmoothingParametersForMarginals(params,100,1.0); NesterovAcceleratedGradient<GraphicalModelType,Minimizer> solver(gm,params); solver.infer();

GraphicalModelType::IndependentFactorType out; for (size_t i=0;i<gm.numberOfVariables();++i) { solver.marginal(i,out_nest); .../* do with the 'out' marginals what you want */ }

/* With ADSal: */ ADSal<GraphicalModelType,Minimizer>::Parameter params; setSmoothingParametersForMarginals(params,100,1.0); ADSal<GraphicalModelType,Minimizer> solver(gm,params); solver.infer(); GraphicalModelType::IndependentFactorType out; for (size_t i=0;i<gm.numberOfVariables();++i) { solver.marginal(i,out); .../* do with the 'out' marginals what you want */ }

Corresponding author: Bogdan Savchynskyy

Definition at line 59 of file smoothing_to_marginals.hxx.